On Wed, 2010-06-09 at 01:03 -0700, nfo wrote:
Could be because you have multiple network interfaces (parallels?
Try adding the local IP you want to use in each config/
This seems to be a frequent issue in recent releases. The problem seems
to arise as follows:
The server has multiple IP addresses, eg:
1: lo: <LOOPBACK,UP,LOWER_UP> mtu 16436 qdisc noqueue state UNKNOWN
link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
inet 127.0.0.1/8 brd 127.255.255.255 scope host lo
inet 127.0.0.2/8 brd 127.255.255.255 scope host secondary lo
2: eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast state UP qlen 1000
link/ether d8:d3:85:a3:33:a4 brd ff:ff:ff:ff:ff:ff
inet 192.168.10.50/24 brd 192.168.10.255 scope global eth0
3: eth1: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast state UP qlen 1000
link/ether d8:d3:85:a3:33:a6 brd ff:ff:ff:ff:ff:ff
inet 192.168.50.50/24 brd 192.168.50.255 scope global eth1
The default config in ES is to bind to 0.0.0.0, ie all addresses, so
in this case it binds to localhost, 192.168.10.50 and 192.168.50.50
Then ES has to choose ONE address to use as the "publish" address, which
is the address that other nodes choose to connect to. In this case it
happens to choose 192.168.50.50
Another node is started, gets the .50.50 address, tries to connect
and because its default gateway is 192.168.10.1, it fails
The solution is to specify either which IP address it should bind to, or
which IP address it should use as the publish address, eg:
The latter method has the advantage that ES still binds to localhost as
well, but is providing the correct address to other nodes.
An alternative is to say:
which binds to the first ethernet interface.
I'm wondering if there should also be the option:
which would only bind to an IP address that falls into that subnet?